Feature ranking


(Erlend Sogge Heggen) #1

So I’ve been giving some thought to how a secondary interest-metric might work. While the UX of the “Super Like” was terrible, I think the original intent of the system had merit. I strongly disagree with the UserVoice voting model:

I’ve thought of a different approach to gaining insight on feature priority:

Feature ranking

This change would add a ranking to your personal votes list once you’ve voted on 2 or more items:

To change the priority of an item, simply drag it up or down in the list (see the arrow).

This would be interesting data. Let’s say you can see “avg ranking” next to the vote count in the topic view. Example:

  • Steam Login
    500 votes
    avg ranking: #21

  • GitLab Login:
    220 votes
    avg ranking: #2

More people are showing interest in “Steam Login”, but almost everyone who wants the “GitLab Login” really wants it.

All your items have ranking:null until you’ve reordered an item for the first time.


(James Kiesel) #2

Sounds a heck of a lot like Instant-Runoff / Preferential voting, while the previous idea sounds a lot like dot voting.

(We’ve gotten tons of interest in dot voting, and have implemented it in Loomio. Instant run-off tends to be slightly less easy-to-understand and thus slightly more misused in my view.)


(Jeff Atwood) #3

I have a different reaction to this. If people are “wasting” their votes on silly stuff like Steam game ownership (which is, really, a very silly proposal) then they have too many votes to give. Reduce the number of votes to just 3 or 2.


(Republic Wireless) #4

I was also thinking dot voting. Instead of putting it in order, just pick the 1, 2 or 3 (maybe set by trust level) that you just can’t do without. It puts a little extra mojo on a smaller subset of your votes. Your limited list of favorites get a force multiplier. It will really help us understand what people feel strongly about.

Maybe tl0 can’t vote, tl1 can vote, tl2 can pick a favorite and tl3 can pick 2 favorites.


(Sam Saffron) #5

My reaction here is that I want to add

This solves the problem described and follows an existing well established paradigm.


(Jeff Atwood) #6

Paradigm where? In what voting process is it ok to cast multiple votes?


(Sam Saffron) #7

User voice has been doing this for about 6 years

Ranked voting is common in politics, but gets very complicated


(Patrick Klug) #8

if you feel more comfortable call it a weighted vote. it’s simply a mechanism to allow the user to say ‘this is important/more important/most important to me’.


(Jeff Atwood) #9

Ranked votes makes more sense. I vote for this first, this second, etc.


(Patrick Klug) #10

it’s not the same though. allowing multiple votes while limiting votes creates a vote-economy (votes are currency). I can spend lots of my votes on a few important issues or I can vote on a bunch of smaller issues equally. It’s the more flexible system since not all users want to prioritize.


(Jeff Atwood) #11

I have zero interest in implenting that system.


(Sam Saffron) #12

To be honest I have zero interest in ranked voting or supervoting and mild interest in “multivoting”

I feel that all 3 systems are solving a problem that largely does not exist. And definitely does not exist in any of the current sites we host including the most popular voting category we have in the wild.

Features - Infinite Flight Community

It is pretty clear what the community want in the above list. What is unclear is what “the powers that be” plan to do about these ideas? What is planned? what is rejected? What is the “official” response?

The issue “supervotes”, “vote priority” and “multivoting” are trying to solve is an ability to get “more signal” from a noisy idea list. Why work on feature X if its everyone’s 3rd pick?

I feel “supervotes” fails hugely in that it is super confusing for end users, how do they decide if they “super” want something vs “want”?

I feel “vote priority” fails hugely cause we will make a fancy page that almost nobody will go to. Even if by some miracle everyone went to said page and diligently ranked every vote, what would you even do with this signal. I feel the ergonomics for vote priority are just bad and I can not think of any way of making the ergonomics good. If you force this decision upfront its a huge mess, if you defer it nobody will bother.

I am partial to “multivoting” cause it makes migrating data from uservoice easier, and its a simpler system for users to comprehend and use. However, it also has some huge flaws, cause the number you see on a topic does not represent number of users, so it can be very misleading. I am also unclear if it will do a better job at ranking ideas vs the current system.

My call here is to shelf multivoting, feature ranking and supervoting till many more of the core issues are sorted with the plugin and there are some examples in the wild of places that need this.


(Jeff Atwood) #13

I agree with all that, however, simply doing it in the actual order you placed the votes is quite straightforward … but it is implicit. You need to make sure you vote for the thing you REALLY want first, then the next thing you want a little bit less, and so on.

It is a shame because we would still need to offer a way to re-order your votes in the UI if you do not cast your most important vote first.


(Barry van Oudtshoorn) #14

Apologies for reviving a relatively old topic on this.

I don’t believe that this would work in the wild – at least for us. Our Suggestions category is open for new posts – and people come up with new ideas all the time, and then want to vote on them. And you may not have realised just how important something is until someone else suggests it. :wink:

This is something that I’ve tried to address in our community by using tags:

This system has worked well (with a few caveats, as I outline in that post). It’d be great to see some of the issues there resolved; to my mind, these are some of the core issues that @sam mentioned that need to be resolved before looking at “nice-to-haves” like multivoting or ranked voting.


(Jim Whitescarver) #15

Group rank ordering in loomio and surveymonkey are good but I hate to introduce another tool just for rank ordering. Without ranking ordering in discourse voting is rather lame. If ranking was included our communities might choose discourse because of its voting prowness.

Ranked choices have great benefits in many situations and may improve democracy.